Hello,
On Wed, 26 Apr 2023 16:32:28 +0900
Yuichiro NAITO <[email protected]> wrote:
> These 2 revisions of 'attr_val' and 'attr_thr' are different on this
> disk.
> The comment says that it's wrong vendor implementation but I can see
> 'smartctl' shows the attributes as follows. NetBSD's atactl doesn't
> see these 2 revisions are same but checks each checksum is valid.
The diff seems correct.
ok?
> diff --git a/sbin/atactl/atactl.c b/sbin/atactl/atactl.c
> index 85dfced8c9a..1f77460ce3d 100644
> --- a/sbin/atactl/atactl.c
> +++ b/sbin/atactl/atactl.c
> @@ -1657,13 +1657,11 @@ device_attr(int argc, char *argv[])
> req.datalen = sizeof(attr_thr);
> ata_command(&req);
>
> - if (attr_val.revision != attr_thr.revision) {
> - /*
> - * Non standard vendor implementation.
> - * Return, since we don't know how to use this.
> - */
> - return;
> - }
> + if (smart_cksum((u_int8_t *)&attr_val, sizeof(attr_val)) != 0)
> + errx(1, "Checksum mismatch (attr_val)");
> +
> + if (smart_cksum((u_int8_t *)&attr_thr, sizeof(attr_thr)) != 0)
> + errx(1, "Checksum mismatch (attr_thr)");
>
> attr = attr_val.attribute;
> thr = attr_thr.threshold;
>
